System self-awakens after powerdown
 

Slackware 14.2 64-bit, ASUS Z170-K, AMI BIOS 3805 05/16/2018, Quad core Intel Core i5-6400, 16 GB DD4 RAM.

I have looked around the web to no avail. As this is solely a Slackware system I am hoping for some ideas on this problem.

After powering down the system, about 7 seconds later, sometimes the system will self-awaken. I am unable to find a cause or common pattern. I might go several days without the system powering on by itself and might go two days in a row where that happens. Doesn't seem to matter the time of day. I have seen the system power on in the day, evening, night.

I have fiddled in the BIOS with ErP Ready. I have tried Disabled, S4+S5, and S5. I don't like s5 because then I can't use the RTC to awaken the system at desired times.

Power On by PCI-E/PCI is enabled. That always succeeds regardless of the ErP Ready config.

I do use the RTC to awaken the system. The default awaken time is 1:01 PM daily. The wakeup time is modified when I record old movies.

At the moment seems I have to conjur up some kind of boot test so I can ensure the system remains powered down. For example, I might record an old movie at 3AM and the system will power off at the end of the recording. Except sometimes the system powers right back on about 7 seconds later. I verify this behavior in the logs the next day.

I have yet to see the system self-awaken after a second power down.

Thanks for any ideas. :)

A faulty power switch? What happens after you press the Power button for at least 8 seconds?

Or marginal power supply.

I've had small form factor Asus motherboards with similar bios hooks and they can be quirky. Try also resetting to default or 'optimized defaults'.

I suppose in some cases the PSU could be at fault but my money is on something in BIOS/EFI. FWIW having spent most of my life in Electronics I am painfully aware and experienced in the importance of any Power Supply, an area that many OEMs use as a cost-cutting favorite. I build my own systems and always buy top of the line PSUs. My voltages are rock solid under the heaviest loads and vary from spec less than 1%, yet sometimes my system will self restart unless I hold the power button down more than 4 seconds twice.

PSU: Seasonic S12II 430B

With respect to load:

16 GB PC-17000 DDR4 (8 + 8 GB) RAM
On-board video: HD Graphics 530
On-board 1 Gbps NIC
On-board sound
WD WD2003FZEX-00Z4SA0 Black SATA-III 2 TB hard drive
WD WD1002FAEX-00Y9A0 Black SATA-III 1 TB hard drive
Mitsumi FA404M BLK multi-card reader
Lite-On iHAS324-08 24X SATA DVD burner
Hauppauge HVR-1600 PCI MCE-Kit TV capture card
Hauppauge HVR-1250 PCIe TV capture card

The last time I checked with my Kill-a-Watt:

79 watts at max CPU frequency
54 watts with CPU idle

BIOS is the latest as of a few weeks ago.

Good thought, but I never power down that way. :) I execute /sbin/shutdown -h now. Likewise in my scripts that power down the system. Nonetheless, later this evening I will try the switch to see if I can repeat the behavior.

This very same thing happened to us on a desktop computer and it finally was traced to a dying CMOS battery. No kidding. Changed the battery and all was well

Okeydokey -- on my to-do list. :)

Does the behavior continue after removing the power source for at least 30 minutes?

Shut down the system, then yank the power cord from the wall before it has a chance to wake up again. Wait 30 minutes, then plug the power cord back into the mains. Watch it for a few minutes, but touch no buttons.

No wake up means a thermal fault; letting the system cool fixed the problem, at least temporarily.

But waking up, even after the system relaxes? ... I got nothing ...

I have not tried that. I use conky to monitor CPU, MB, and HD temps. CPU and MB temps stay low at about 30 C. HDs at about 40-43 C.

That said, perhaps a stray bit somewhere is hanging, or something like that. A full switch power off is worth a shot.

The CMOS battery looks good at 3.16 volts, but I changed anyway.

With troubleshooting, always start with the easiest. The first things I would check is to make sure your network and none of your peripherals are waking it up. Unplug your keyboard and mouse and your network cable... maybe even extra devices like external harddrives or a printer. Then you can go into the case and remove any unnecessary PCIe cards and your power cable connector to the motherboard (to verify the power switch isn't faulty). After that, if you have an extra PSU, give that a try.

Hopefully this can help narrow down the problem...

As I've already mentioned my system only refuses to stay down after a software call to shutdown as in either a direct command like "shutdown -h now" or from KDE "Leave > Shutdown" sometimes... maybe 40% of the time. It's hard to have an exact figure because I only rarely do a hard shutdown. My system is almost always on and reboots are far more common than shutdowns because I have several operating systems on this box, one, not my main, is a dedicated DAW.

That someone mentioned a weak CMOS battery only strengthens my suspicion that it has to do with BIOS/EFI, either a setting or a minor problem in the BIOS/Efi code. IIRC there is only a single bit difference between hard shutdown and reboot which is reflected in the difference between "shutdown -r now" and "shutdown -h now". Additionally there are several "wake" functions in BIOS/Efi that are also single bit toggles. If I was concerned about this rare and minor problem (for me, anyway) my first choice would be to flash the BIOS and monitor it's effect.

I shared in a previous post that the "BIOS is the latest as of a few weeks ago." :)

Since my last post I have had Power On by PCI-E/PCI in the BIOS disabled. The system has not self-awakened once.

Looks like the next step is deciding whether this is a BIOS bug or an actual device is causing the wake trigger.

Although the BIOS option is named Power On by PCI-E/PCI, this is the option that controls WOL magic packets. I verified this as I no longer can power on the computer with a WOL magic packet.

Is there an actual WOL signal being transmitted somehow, somewhere? When I power down this system the only other live device on the network is the router. An easy test is powering off the router before powering down the computer. I strongly doubt the router and if proven correct, that leaves something internal to the computer. As the BIOS option refers to PCI-E/PCI, one test is to pull all internal cards and test for self-awakening. Inconvenient but I see no other way to test.

I have read that some people see this same behavior with USB devices. I guess then repeat the power downs with all USB devices disconnected. I'll use a PS/2 mouse for a while.

https://wiki.archlinux.org/index.php/Wake-on-LAN

with results like
When WakeOnLAN is enabled, anything else than "g" will make the computer to power up at first network activity, just like your computer do.

I am aware of that. My system is set to g only. On shutdown I run ethtool -s eth0 wol g. Been doing that for a long time. Thanks. :)